INDEX

Explanations

names of individuals, particularly those starting with the letter "G" and "H"

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

awaru

-0.92

Discussion

-0.76

egal

-0.75

kat

-0.75

wrapper

-0.74

kers

-0.73

ked

-0.72

anova

-0.72

atile

-0.72

adin

-0.71

POSITIVE LOGITS

 Griffith

0.89

rious

0.85

 Kaufman

0.77

 Berger

0.75

 Friedman

0.75

 Chandler

0.74

 Herbert

0.74

 Cheong

0.74

 Burr

0.73

 Ford

0.73

Activations Density 0.073%

names of individuals, particularly those starting with the letter "G" and "H"

No Comments

No Known Activations